1: %\documentstyle[prl,aps,preprint,psfig]{revtex}

2: %\documentclass[11pt]{article}

3: 

4: \documentclass[floats,aps,showpacs,twocolumn]{revtex4}

5: 

6: \usepackage[dvips]{graphicx}

7: \usepackage{amsfonts}

8: 

9: %\newcommand{\RightText}{

10: %%{\it Revision: 0.1}, 13.09.04 % PL main version

11: %%{\it Revision: 0.2}, 01.10.04 % AM suggested changes

12: %%{\it Revision: 0.3}, 19.10.04 % PL, AR changes

13: %{\it Revision: 0.4}, 21.10.04 % AM minor changes

14: %}

15: 

16: %\markright{\RightText}

17: 

18: \newcommand{\PSImagx}[2]{\includegraphics[width=#2]{#1}} % command from Arnd

19: 

20: \newcommand{\tauh}{\tau_{\scriptscriptstyle H}}

21: \newcommand{\tmin}{\tau_{\text{min}}}

22: \newcommand{\xmin}{x_{\text{min}}}

23: \newcommand{\leqa}{\stackrel{<}{\scriptstyle \sim}}

24: \newcommand{\geqa}{\stackrel{>}{\scriptstyle \sim}}

25: \newcommand{\kf}{k_{\scriptscriptstyle F}}

26: \newcommand{\ef}{\epsilon_{\text {\tiny F}}}

27: \newcommand{\ecb}{\overline{\cal E}}

28: \newcommand{\ect}{\widetilde{\cal E}}

29: \newcommand{\ob}{\overline{\Omega}}

30: \newcommand{\ot}{\widetilde{\Omega}}

31: \newcommand{\rmb}{\rho_{\text {\tiny MB}}}

32: \newcommand{\rhob}{\overline{\rho}}

33: \newcommand{\rhot}{\widetilde{\rho}}

34: \newcommand{\ec}{{\cal E}}

35: \newcommand{\sca}{{\cal S}}

36: \newcommand{\scab}{\overline{{\cal S}}}

37: \newcommand{\scat}{\widetilde{{\cal S}}}

38: \newcommand{\dc}{{\cal D}}

39: \newcommand{\nc}{{\cal N}}

40: \newcommand{\ncb}{\overline{\cal N}}

41: \newcommand{\nct}{\widetilde{\cal N}}

42: \newcommand{\mub}{\overline{\mu}}

43: \newcommand{\mut}{\widetilde{\mu}}

44: \newcommand{\tb}{\overline{T}}

45: \newcommand{\tti}{\widetilde{T}}

46: \newcommand{\taut}{\tau_{\scriptscriptstyle T}}

47: \newcommand{\xh}{x_{\text{H}}}

48: 

49: \newlength{\figwidth}

50: \setcounter{equation}{0}

51: \setlength{\figwidth}{3.0in}

52: 

53: 

54: \begin{document}

55: 

56: \title{Fluctuations in the level density of a Fermi gas}

57: 

58: \author{P. Leboeuf$^1$, A. G. Monastra$^2$, and A. Rela\~no$^3$}

59: 

60: \affiliation{

61: $^1$Laboratoire de Physique Th\'eorique et Mod\`eles Statistiques,

62: B\^at. 100, \\ Universit\'e de Paris-Sud, 91405 Orsay Cedex, France\\

63: $^2$Institut f\"ur Theoretische Physik, Technische

64: Universit\"at Dresden, 01062 Dresden, Germany\\

65: $^3$Departamento de F\'{\i}sica At\'omica, Molecular y Nuclear,

66: Universidad Complutense de Madrid, E-28040 Madrid, Spain}

67: 

68: %\date{\today, \RightText}

69: \date{\today}

70: 

71: \begin{abstract} 

72: We present a theory that accurately describes the counting of excited states

73: of a noninteracting fermionic gas. At high excitation energies the results

74: reproduce Bethe's theory. At low energies oscillatory corrections to the

75: many--body density of states, related to shell effects, are obtained. The

76: fluctuations depend non-trivially on energy and particle number. Universality

77: and connections with Poisson statistics and random matrix theory are

78: established for regular and chaotic single--particle motion.

79: \end{abstract}

80: 

81: \pacs{PACS numbers: 21.10.Ma, 24.60.-k, 05.45.Mt}

82: 

83: \maketitle

84: 

85: The level density is a characteristic property of every many--body

86: quantum mechanical system. Its precise determination is often a key

87: ingredient in the calculation of different processes, like compound

88: nuclear decay rates, yields of evaporation residues to populate exotic

89: nuclei, or thermonuclear rates in astrophysical processes. The first

90: and main step towards the understanding of the density was given by

91: H. Bethe more than sixty years ago. He considered $A$ noninteracting

92: fermions moving in a mean--field potential. By this model he showed

93: that the number of excited states of the many--body (MB) system

94: contained in a small energy window $dQ$ at energy $Q$, $\rmb (A,Q)

95: dQ$, is \cite{bethe}

96: \begin{equation} \label{bethe}

97: \rmb (A,Q) = \frac{1}{\sqrt{48}\ Q} \exp \sqrt{\frac{2}{3}\pi^2 \rhob \ Q} \ .

98: \end{equation}

99: Here $\rhob$ is the average single--particle (SP) density of states at Fermi

100: energy $\ef$, and $Q$ is measured with respect to the ground state energy. For

101: a given potential, $\rhob$ is in general a function of $A$. This expression

102: assumes that the excitation energy is, on the one hand, large compared to the

103: SP mean level spacing $\delta = \rhob^{-1}$ and, on the other, small compared

104: to $\ef$ (degenerate gas approximation). Most of the present knowledge

105: concerning the density of states is based on refinements of Bethe's result.

106: 

107: In Eq.~(\ref{bethe}) all the information concerning the SP spectrum is encoded

108: in a unique parameter, $\rhob$, that describes its average behavior. The exact

109: dependence of $\rmb$ on $A$ and $Q$ is sensitive, however, to the detailed

110: arrangement of the SP energy levels around the Fermi energy. After Bethe's

111: work, more accurate calculations of $\rmb$ were made \cite{bloch}. Schematic

112: shell corrections related to a periodic fluctuation of the SP density were

113: computed in \cite{ros} (see also \cite{bm}), introducing the so--called

114: backshifted Bethe formula. Several modifications of Eq.(\ref{bethe}) have been

115: done to match the experimental results. These models take into account, for

116: example, shell effects, pairing corrections and residual interactions

117: \cite{para}, introducing a multitude of coexisting phenomenological

118: parameterizations. The present status of the understanding do not allow to

119: draw a clear theoretical picture of the functional dependence of $\rmb$ with

120: $A$ and $Q$.

121: 

122: Within an independent particle model we demonstrate, through a general theory,

123: the role played by shell effects in the MB density of states. Superimposed to

124: the smooth growth of the density described by Eq.(\ref{bethe}), there are

125: oscillatory corrections. To lowest order, these corrections are directly

126: related to energy fluctuations of the system, that may be expressed in terms

127: of sums over the classical periodic orbits. We make a statistical analysis of

128: the density fluctuations, in particular of their probability distribution,

129: typical size, and temperature dependence \cite{note}. We show that the

130: amplitude of the shell corrections grows linearly at low temperatures, and

131: decays as $T^{-1}$ when $T \gg E_c /2\pi^2$, where $E_c$ is the energy

132: conjugate to the time of flight across the system (temperature is measured in

133: units of Boltzmann constant). The statistical properties of the density

134: fluctuations strongly depend on the regular or chaotic nature of the

135: underlying classical SP motion. Universal properties for each class (regular

136: or chaotic) are found for $T \ll E_c /2\pi^2$. Besides the fluctuations, we

137: find and discuss corrections to the smooth part of the density in the exponent

138: of Eq.(\ref{bethe}). The present analysis considers general properties of

139: fermionic systems. Applications to the specific problem of the nuclear density

140: and to the interpretation of the experimental results will be done in a

141: separate contribution.

142: 

143: The MB density of states is defined as

144: \begin{equation} \label{rhodef}

145: \rmb (A,E) = \sum_N \sum_j \delta (A- N) \ \delta \left( E - E_{N,j} \right) \ ,

146: \end{equation}

147: where $E_{N,j}=\sum_i n_{j,i} \epsilon_i$ is the energy of the $j$-th SP

148: configuration of $N = \sum_i n_{j,i}$ particles, $\epsilon_i$ the SP energies,

149: and $n_{j,i}=0,1$ are the occupation numbers.

150: 

151: When the SP spectrum consists of equidistant levels separated by

152: $\delta$, e.g. the spectrum of a one dimensional harmonic oscillator,

153: an exact answer to the counting problem exists. The MB excitation

154: energies are integer multiples of $\delta$, each level having a

155: nontrivial degeneracy. It is easy to see that, for a given level

156: defined by an integer, the degeneracy is the same as the number of

157: ways into which the integer can be decomposed as a sum of integers

158: (the partition number). An asymptotic approximation to this

159: well--known mathematical problem was obtained by Hardy and Ramanujan,

160: and later on Rademacher found a convergent series

161: \cite{hrr}. Expressing it in terms of an expansion in terms of $\rhob

162: Q$, the result reads

163: \begin{eqnarray} \label{shr2}

164: && \log \left( \rmb^{\text {\tiny HO}}/\rhob \right) = \sqrt{ \frac{2}{3}

165: \pi^2 \rhob \ Q } - \log \left( \sqrt{48} \ \rhob Q \right) - \\ && \ \

166: \frac{\pi^2 + 72}{24 \sqrt{6} \pi} (\rhob Q)^{-\frac{1}{2}} - \left(\frac{3}{4

167: \pi^2}- \frac{1}{24} \right) (\rhob Q)^{-1} + {\cal O} ((\rhob

168: Q)^{-\frac{3}{2}}) \nonumber \ .

169: \end{eqnarray}

170: The first two terms of this expansion reproduce Bethe's formula.

171: 

172: For an arbitrary SP spectrum the computation of the density of excited

173: states of a fermionic system is a difficult combinatorial problem for

174: which no exact solution is known. We expect, however, that

175: Eq. (\ref{shr2}) gives correctly the contribution of the average part

176: of the SP spectrum. What we are seeking here are the variations of the

177: MB density due to fluctuations of the SP spectrum with respect to a

178: perfectly ordered spectrum. A convenient way to express an approximate

179: solution is by means of an inverse Laplace transform of

180: Eq.(\ref{rhodef}). A saddle point approximation of the resulting

181: integrals yields \cite{bm,fow}

182: \begin{equation} \label{rhosp}

183: \rmb (A,E) = {\rm e}^{\sca (\mu,T)}/ \ 2\pi \sqrt{|\dc (\mu,T)|} \ ,

184: \end{equation}

185: valid in the degenerate gas approximation. The dependence on $A$ and $E$ in

186: Eq.~(\ref{rhosp}) arises from the saddle point conditions that fix the value

187: of the chemical potential $\mu$ and temperature $T$ of the gas

188: \begin{equation} \label{spc}

189: \nc (\mu, T) = A \ , \;\;\;\;\;\;\;\;\;\;\;\;\;\;\; \ec (\mu, T) = E \ . 

190: \end{equation}

191: The functions $\nc=-\partial \Omega/\partial \mu|_{\scriptscriptstyle T}$ and 

192: \begin{equation} \label{u}

193: \ec (\mu,T) = \Omega (\mu,T) + \mu \nc (\mu,T) + T \sca (\mu,T)

194: \end{equation}

195: are the particle number and energy functions of the gas, respectively,

196: with the entropy $\sca (\mu,T) =-\partial \Omega/\partial T|_\mu$ and

197: the grand potential $\Omega = - T \int d\epsilon \rho (\epsilon) \log

198: \left[1+{\rm e}^{(\mu-\epsilon)/T}\right]$, where $\rho (\epsilon) =

199: \sum_i \delta (\epsilon-\epsilon_i)$ is the SP density of states. In

200: terms of the previous functions, the determinant in Eq.(\ref{rhosp})

201: is defined as $\dc (\mu,T) = T^3 \left[ \partial \nc/\partial

202: \mu|_{\scriptscriptstyle T} \

203: \partial \ec/\partial T|_\mu - \partial \nc/\partial T|_\mu \ \partial

204: \ec/\partial \mu|_{\scriptscriptstyle T} \right]$. All the necessary quantities 

205: involved in the computation of $\rmb$ are therefore defined from

206: $\Omega (\mu,T)$.

207: 

208: The task consists in the evaluation of the entropy and the determinant at

209: values of $\mu$ and $T$ that satisfy the conditions (\ref{spc}). As $A$ and

210: $E$ are varied in Eqs.(\ref{spc}), in general $\mu$ and $T$ do not have a

211: smooth and gentle behavior, because the functions $\nc$ and $\ec$ may have

212: sudden changes due to the discrete nature of the SP spectrum. The difficulty

213: lies in a proper treatment of the smooth part of the variations as well as the

214: fluctuations with respect to it. To this purpose it is convenient to use the

215: semiclassical approximation \cite{gutz}. The SP density of states is

216: decomposed into the sum of a smooth term ($\rhob$, given by a Thomas Fermi

217: approximation or Weyl series) plus oscillatory terms $(\rhot)$, $\rho = \rhob

218: + \rhot$. Taking into account only the smooth part of the SP density of

219: states, the calculation of $\rmb$ is relatively straightforward and leads to

220: Bethe's formula. The oscillatory part depends on the primitive classical

221: periodic orbits $p$ (and their repetitions $r$), $\rhot = 2 \sum_p

222: \sum_{r=1}^{\infty} A_{p,r} \cos \left[ r S_p/\hbar+ \nu_{p,r} \right]$. Each

223: orbit is characterized by its action $S_p$, stability amplitude $A_{p,r}$, and

224: Maslov index $\nu_{p,r}$. When inserted into the definition of the grand

225: potential, and after integration with respect to the energy, the fluctuating

226: part of $\Omega$ is given by \cite{ruj}

227: \begin{equation} \label{grandosc}

228: \widetilde{\Omega} (\mu,T)= 2 \hbar^2 \sum_p \sum_{r=1}^{\infty} \frac{A_{p,r}

229: ~ \kappa ( \frac{r ~ \tau_p}{\taut})}{r^2 ~ \tau_p^2} \cos \left(

230: \frac{\scriptstyle r S_p}{\scriptstyle \hbar}+ \nu_{p,r} \right).

231: \end{equation}

232: Here $\tau_p$ is the period of the periodic orbit, and $\kappa (x) = x/\sinh

233: (x) $ is a temperature factor that introduces the time scale $\taut =

234: \hbar/(\pi T)$ conjugate to the temperature. This expression describes the

235: departures of $\Omega$ with respect to its mean behavior due to the

236: fluctuations of the SP spectrum. By simple derivation of the grand potential

237: the smooth and fluctuating part of all other thermodynamic functions can be

238: obtained. Intensive functions like the chemical potential and the temperature

239: should also be decomposed. We define their smooth parts by the implicit

240: equations

241: \begin{equation} \label{spc2}

242: \ncb (\mub, \tb) = A \ , \;\;\;\;\;\;\;\;\;\;\;\;\;\;\; \ecb (\mub,

243: \tb) = E \ ,

244: \end{equation}

245: whereas the fluctuating parts are $\mut = \mu - \mub$ and $\tti = T -

246: \tb$.

247: 

248: Using Eqs.(\ref{spc}) and the decomposition of the different

249: functions, Eq.(\ref{u}) may be rewritten as

250: \begin{equation} \label{s1}

251: (\tb + \tti) \sca = E - \ob (\mu,T) - \ot (\mu,T) - (\mub + \mut) A \ .

252: \end{equation}

253: If $A$ is large, the smooth part of the grand potential can be expanded in

254: power series of $\mut$ and $\tti$. Keeping terms up to first order, using that

255: in the degenerate gas approximation $\ob (\mub,\tb) \simeq \ob (\mub,0) -

256: (\pi^2/6) \rhob \ \tb^2$, and neglecting variations of $\rhob$, Eq.(\ref{s1})

257: becomes

258: \begin{eqnarray} \label{s2}

259: (\tb + \tti) \sca &=& E - \ob (\mub,0) + \frac{\pi^2}{6} \rhob \ \tb^2 +

260: \overline{\sca} (\mub,\tb) \ \tti - \nonumber \\ && \ot (\mu,T) - \mub A \ .

261: \end{eqnarray}

262: This equation is valid for an arbitrary temperature. At zero temperature, the

263: energy of the system is the ground state energy $E_0$, and Eq.(\ref{s2})

264: reduces to

265: \begin{equation} \label{s3}

266: 0 = E_0 - \ob (\mub,0) - \ot (\mu_0,0) - \mub A \ .

267: \end{equation}

268: $\mub$ remains constant if variations of $\rhob$ are neglected, but the

269: fluctuating part still depends on temperature. As a consequence the chemical

270: potentials are slightly different. The idea is to subtract Eq.(\ref{s3}) from

271: (\ref{s2}). Before doing that we first need to properly analyze some of the

272: terms. The difference $E-E_0=Q$ is the excitation energy of the gas, as well

273: as the term $(\pi^2/6) \rhob \ \tb^2 =Q$. The latter equation follows from the

274: dependence of the energy on temperature in the degenerate gas approximation,

275: $\ecb (\mub,\tb) \simeq \ecb (\mub,0) + (\pi^2/6) \rhob \ \tb^2$, and the

276: stationary phase condition (\ref{spc2}) for $\ecb$ (and the corresponding one

277: at zero temperature). It allows to express the temperature in terms of $Q$.

278: Similarly, $\mub$ is obtained by inversion of $\ncb (\mub,\tb)$ in

279: (\ref{spc2}) which, in the degenerate gas approximation and neglecting

280: variations of $\rhob$, is independent of $\tb$.

281: 

282: Returning to Eq.(\ref{s2}), the term $\tti (\sca - \overline{\sca} (\mub,\tb))

283: = \tti \widetilde{\sca}$ is of second order, and is therefore neglected.

284: After subtraction of Eq.(\ref{s3}) and expressing $\tb$ in terms of $Q$, the

285: entropy may be expressed as

286: \begin{eqnarray} \label{sdef}

287: \sca (A,Q) &=& \sqrt{\frac{2}{3} \pi^2 \rhob \ Q} + \frac{1}{\tb}

288: \left[ \ot (\mub,0) - \ot (\mub,\tb) \right] \\

289: &\approx& \sqrt{\frac{2}{3} \pi^2 \rhob \ \left[ Q + \ot (\mub,0) -

290: \ot (\mub,\tb) \right]} \ , \label{sdef2}

291: \end{eqnarray}

292: where, to lowest order, $\ot$ has been evaluated at $(\mub, \tb)$ (a similar

293: though improved accuracy is obtained using $\mu$ and $T$, cf. Eq.(\ref{hro})

294: below). Equation (\ref{sdef2}) is valid if $\ot (\mub,0) - \ot (\mub,\tb) \ll

295: Q$, and puts the result under the form of a backshifted Bethe formula.

296: 

297: A more physical interpretation of Eqs.(\ref{sdef}) and (\ref{sdef2}), that

298: expresses the result in an entirely microcanonical language, is provided by

299: the following connection. It has been shown previously \cite{lm3,sev} that, to

300: leading order in an expansion in terms of $\mut$, $\ot (\mub,\tb) = \ect

301: (A,Q)$, where $\ect (A,Q)$ are the (shell) fluctuations of the energy of the

302: gas at a fixed number of particles and excitation energy. The shell

303: corrections of Eq.(\ref{sdef2}) are therefore directly related to the

304: fluctuations of the energy of the system.

305: 

306: A similar analysis can be performed for the determinant $\dc (\mu,T)$ in

307: Eq.(\ref{rhosp}). To leading order we find

308: \begin{equation} \label{det}

309: \dc = \frac{\pi^2}{3} \rhob^2 \ \tb^4 = \frac{12}{\pi^2} \ Q^2 \ ,

310: \end{equation}

311: recovering the same form obtained by Bethe ($2\pi \sqrt{\dc}=\sqrt{48}

312: \ Q$, see Eq.(\ref{bethe})). There exist oscillatory corrections to

313: $\dc$, but these are exponentially small compared to those associated

314: with Eq.(\ref{sdef}), and we neglect them.

315: 

316: Equation (\ref{sdef}) is the central result of the paper. It expresses,

317: together with Eq.(\ref{det}), the MB density of states in terms of the

318: particle number $A$ and excitation energy $Q$. It decomposes $\sca$ into a

319: smooth and a fluctuating part, $\sca = \scab + \scat$. The smooth part is

320: given by the usual Bethe's result (corrections to it may be obtained by

321: keeping higher order terms in $\mut$ and $\tti$ in the expansion). The novelty

322: in Eq.(\ref{sdef}) is the additional contribution of oscillatory corrections,

323: $\scat = [\ot (\mub,0) - \ot (\mub,\tb)]/\tb$. They describe shell corrections

324: to the MB density. $\widetilde{\Omega}$ depends on temperature only through

325: the function $\kappa$. The effect of this function is to exponentially

326: suppress the contribution of periodic orbits whose period $\tau_p \gg \taut$

327: \cite{ruj,lm3}. Since there is no suppression at $T=0$ (because $\taut

328: \rightarrow \infty$), only orbits whose period $\tau_p \gtrsim \taut$

329: contribute to the difference $\ot (\mub,0) - \ot (\mub,\tb)$. At temperatures

330: such that $\taut \ll \tmin$, where $\tmin$ is the period of the shortest

331: periodic orbit of the system, the term $\ot (\mub,\tb)$ becomes exponentially

332: small, and only $\ot (\mub,0)$ remains. The shell correction $\scat$ therefore

333: decays as $\tb^{-1}$. This decay contrasts with the more common exponential

334: damping observed in other thermodynamic quantities \cite{lm3}.

335: 

336: The behavior of $\widetilde{\Omega}$ (and therefore of $\scat$) strongly

337: depends on whether $\mu$ or $T$ are varied. As discussed above, a temperature

338: variation modifies the prefactors of the summands in $\widetilde{\Omega}$

339: (through the function $\kappa$), and therefore produces gentle variations of

340: $\scat$. In contrast, $A_{p,r}$, $\tau_p$ and $S_p$ depend on $\mu$. For large

341: values of $\mu$, $S_p \gg \hbar$ and the dominant variation with the particle

342: number (or any other parameter that modifies the actions) comes from the

343: argument of the cosine function in $\widetilde{\Omega}$. Rapid oscillations of

344: $\scat$ are therefore generically expected when the number of particles is

345: varied.

346: 

347: A clearer picture of how the fluctuations behave may be obtained through a

348: statistical analysis. There are two relevant SP scales in the analysis, the

349: mean spacing $\delta$ and the energy conjugate to the shortest periodic orbit,

350: $E_c = h/\tmin$. The ratio $g = E_c/\delta$ is typically much larger than $1$

351: (for instance, $g\sim A^{2/3}$ in a three dimensional cavity). The most simple

352: property of the fluctuations is $\langle \scat \rangle =0$, where the brackets

353: denote an average over a suitable chemical potential window. This result is

354: valid only to first order in the expansion; it can be shown that higher order

355: terms contribute to a non--zero average. The next non--trivial statistical

356: property is the variance $\langle \scat^2 \rangle$, that may be computed using

357: Eq.(\ref{grandosc}). The result is $\langle \scat^2 \rangle = (1/2)

358: \int_{0}^\infty d x \ K(x,\xh) \left[ 1 - \kappa (x) \right]^2 /x^4$, where

359: $K(x,\xh)$ is the rescaled form factor of the SP spectrum (cf Eq.(36) in

360: Ref.\cite{lm3}). The latter function depends on the rescaled Heisenberg time

361: $\xh=h \rhob/\taut$. It describes system--dependent features for $x$ of the

362: order of $\xmin = \tmin/\taut$, while it is believed to be universal for $x

363: \gg \xmin$. The universality class depends on the regular or chaotic nature of

364: the dynamics, and on its symmetry properties. Taking into account the basic

365: properties of $K(x,\xh)$, in chaotic systems we find three different regimes

366: for $\langle \scat^2 \rangle$ as a function of $\tb$:

367: 

368: \noindent (i) Low temperatures $2\pi^2 \tb \ll \delta$. In this regime

369: $\langle \scat^2 \rangle = c_4 \pi^2 \tb/\delta$, where $c_4 = 0.0609...$.

370: 

371: \noindent (ii) Intermediate temperatures $\delta \ll 2\pi^2 \tb \ll E_c$. In

372: this regime the size of the fluctuations saturate at a universal

373: constant $\langle \scat^2 \rangle = c_3/\beta$, where $c_3 =

374: 0.1023\ldots$ and $\beta = 1 (2)$ for systems with (without)

375: time--reversal invariance.

376: 

377: \noindent (iii) High temperatures $2\pi^2 \tb \gg E_c$. The size of the

378: fluctuations decreases with excitation energy, $\langle \scat^2 \rangle =

379: \langle \ot^2 (\mub,0) \rangle [ 1 - 8 \ {\rm e}^{-2 \pi^2 \tb/E_c} ]/ \tb^2$.

380: After an exponential transient, a power--law decay $\langle \scat^2

381: \rangle^{1/2} \propto \tb^{-1}$ is obtained.

382: 

383: The situation is different in integrable systems, where only two regimes are

384: found. At low temperatures the result is identical to that of chaotic systems.

385: The difference is that now the growth extends up to much higher temperatures,

386: $2 \pi^2 \tb \approx E_c$, without saturation. At that temperature the

387: variance of the fluctuations is of order $g$. In integrable systems, the

388: maximum amplitude of the fluctuations is therefore reached at $2 \pi^2 \tb

389: \approx E_c$, and its typical size is much larger than in chaotic systems. At

390: high temperatures $2 \pi^2 \tb \gg E_c$ the decay is almost identical to that

391: of chaotic systems (the coefficient 8 is replaced by a 12).

392: 

393: As shown elsewhere \cite{lm3}, $\ot (\mub,\tb)$ is dominated, at any $\tb$, by

394: the shortest classical periodic orbits. In contrast, the difference $\ot

395: (\mub,0)-\ot (\mub,\tb)$ depends on orbits whose period $\tau_p \geqa \taut$.

396: For temperatures $2 \pi^2 \tb \ll E_c$ the statistical properties of these

397: orbits are universal, and correspondingly the probability distribution

398: function of $\scat$ is expected to be universal, in the sense that at a given

399: temperature it should only depend on the nature of the underlying classical

400: dynamics (regular or chaotic), and the symmetries of the system. This

401: statement is supported by the fact that in the limit $\tb \rightarrow 0$,

402: $\scat \approx - \partial \ot (\mub,\tb) /\partial \tb$. The probability

403: distribution of the latter quantity was studied in Ref.\cite{lm3}; it was

404: shown that it coincides at low temperatures with that obtained from a Poisson

405: spectrum for integrable systems and from a random matrix spectrum for chaotic

406: ones. As the temperature is raised, the universality of the probability

407: distribution of $\scat$ will be lost for temperatures of the order or greater

408: than $E_c$, where system specific features are revealed.

409: 

410: \begin{figure}

411: \includegraphics[width=8.5cm,height=7.0cm]{lmr_figure.eps}

412: \caption{The logarithm of the many-body density of states of $A$

413:      noninteracting fermions in a rectangular billiard of sides $a = \sqrt{(1

414:      + \sqrt{5})/2}$ and $b=a^{-1}$. Dots: numerical computation at three

415:      different excitation energies; solid curves:  theoretical prediction

416:      (\ref{hro}); dashed curves: smooth part Eq.(\ref{shr2}).}

417: \end{figure}

418: 

419: We have checked some of our predictions by a direct numerical counting of the

420: MB density of states in a particular system. Figure 1 shows the results

421: obtained for a gas of about 2000 fermions contained in a two--dimensional

422: rectangular cavity, an integrable system. For each number of particles we

423: compute the MB density of states at three different temperatures, measured in

424: units of $T_{\delta}=\delta/2 \pi^2$ and $T_c =E_c /2 \pi^2$. The theoretical

425: curve is computed according to the expression

426: \begin{equation} \label{hro}

427: \log ( \rmb/\rhob ) = \log ( \rmb^{\text{\tiny HO}}/\rhob ) + [ \ot

428: (\mu,0) - \ot (\mu,\tb) ]/\tb \ ,

429: \end{equation}

430: where $\ot$ is given by the periodic orbits of the rectangle. Though the

431: results are similar, $\mu$ instead of $\mub$ is used to obtained a more

432: accurate description of the numerical data. A systematic deviation is observed

433: between theory and numerics if the corrections arising from the exact

434: expression Eq.(\ref{shr2}) are not included. Notice the extremely good

435: accuracy of this equation, either for the average value of the density as well

436: as for the fluctuations.

437: 

438: \noindent The Laboratoire de Physique Th\'eorique et Mod\`eles Statistiques is

439: an Unit\'e de recherche de l'Universit\'e Paris XI associ\'ee au CNRS.

440: 

441: \begin{thebibliography}{10}

442: \bibitem{bethe} H. A. Bethe, Phys. Rev. {\bf 50} (1936) 332.

443: \bibitem{bloch} C. Bloch, Phys. Rev. {\bf 93} (1954) 1094.

444: \bibitem{ros} N. Rosenzweig, Phys. Rev. {\bf 108} (1957) 817.

445: \bibitem{bm} A. Bohr and B. R. Mottelson, {\it Nuclear Structure}, Vol.I,

446: Benjamin, Reading, Massachusetts, 1969.

447: \bibitem{para} A. Gilbert and A. G. W. Cameron, Can. J. Phys. {\bf 43} (1965)

448: 1446; A. V. Ignatyuk, G. N. Smirenkin, and A. S. Tishin, Sov. J. Nucl. Phys.

449: {\bf 21} (1975) 255; K. Kataria, V. S. Ramamurthy, and S. S. Kapoor, Phys.

450: Rev. C {\bf 18} (1978) 549; Y. Alhassid, G. F. Bertsch, and L. Fang, Phys.

451: Rev. C {\bf 68} (2003) 044322.

452: \bibitem{note} All over the text, we may either refer to the excitation energy

453: $Q$ of the system or to its temperature $T$.

454: \bibitem{hrr} G. H. Hardy and S. Ramanujan, Proc. London Math. Soc. {\bf 17}

455: (1918) 75; H. Rademacher, Proc. London Math. Soc. {\bf 43} (1937) 241.

456: \bibitem{fow} R. H. Fowler, {\it Statistical Mechanics}, Macmillan Company, New

457: York, 1936.

458: \bibitem{gutz} M. C. Gutzwiller, J. Math. Phys. {\bf 10} (1969) 1004; R. Balian

459: and C. Bloch, Ann. Phys. (N.Y.) {\bf 69} (1972) 76.

460: \bibitem{ruj} K. Richter, D. Ullmo and R. Jalabert, Phys. Rep. {\bf 276}

461: (1996) 1.

462: \bibitem{lm3} P. Leboeuf and A. G. Monastra, Ann. Phys. {\bf 297} (2002) 127.

463: \bibitem{sev} P. Leboeuf, {\it Regularity and chaos in the nuclear masses},

464: Lectures delivered at the VIII Hispalensis International Summer School,

465: Sevilla, Spain, June 2003 (to appear in {\sl Lecture Notes in Physics},

466: Springer--Verlag, Eds. J. M. Arias and M. Lozano); nucl-th/0406064.

467: \end{thebibliography}

468: 

469: \end{document}

470: